Recherche Images Maps Play YouTube Actualités Gmail Drive Plus »
Connexion
Les utilisateurs de lecteurs d'écran peuvent cliquer sur ce lien pour activer le mode d'accessibilité. Celui-ci propose les mêmes fonctionnalités principales, mais il est optimisé pour votre lecteur d'écran.

Brevets

  1. Recherche avancée dans les brevets
Numéro de publicationUS20080276227 A1
Type de publicationDemande
Numéro de demandeUS 11/744,876
Date de publication6 nov. 2008
Date de dépôt6 mai 2007
Date de priorité6 mai 2007
Autre référence de publicationUS8464225
Numéro de publication11744876, 744876, US 2008/0276227 A1, US 2008/276227 A1, US 20080276227 A1, US 20080276227A1, US 2008276227 A1, US 2008276227A1, US-A1-20080276227, US-A1-2008276227, US2008/0276227A1, US2008/276227A1, US20080276227 A1, US20080276227A1, US2008276227 A1, US2008276227A1
InventeursBernd Greifeneder
Cessionnaire d'origineBernd Greifeneder
Exporter la citationBiBTeX, EndNote, RefMan
Liens externes: USPTO, Cession USPTO, Espacenet
Method and System for Adaptive, Generic Code Instrumentation using Run-time or Load-time generated Inheritance Information for Diagnosis and Monitoring Application Performance and Failure
US 20080276227 A1
Résumé
The present system and method provides a system for generic, run-time adaptive placement of bytecode instrumentation, taking object oriented inheritance relationships into account. Said inheritance relationships are stored in an inheritance repository. The inheritance repository that mirrors the structure of the monitored application is created at run-time and updated if the code base of the monitored application changes either dynamically at run-time or by manually changing the configuration and restarting the application. The inheritance repository contains meta-data of application classes and their relationships, like direct and indirect inheritance. The inheritance repository information is used to evaluate generic instrumentation placement rules, like rules that match to methods of classes that inherit from a specific base class. The inheritance repository is generated concurrently to instrumentation placement at application load-time or run-time and is persisted between application runs to enable dedicated adaptation runs to create the repository. Detection of class relationships and instrumentation placement at run-time make the presented system suitable to monitor adaptive applications that dynamically adapt used components, libraries and execution code according to run-time conditions.
Images(10)
Previous page
Next page
Revendications(20)
1) An apparatus for modifying original byte code at runtime, comprising:
an agent intercepting the loading of original byte code by the native loader, and capturing the original byte code and transmitting the original byte code to an instrumentation engine
an instrumentation engine to receive original byte code;
the instrumentation engine converting the original byte code to instrumented byte code and transmitting the instrumented byte code to the agent;
wherein the agent forwards the received instrumented byte code to the runtime environment in a way that the instrumented byte code is loaded instead of the original byte code.
2) An apparatus for modifying original byte code at runtime as in claim 1, wherein the metadata included in byte code is extracted and wherein the extracted metadata is used to evaluate instrumentation meta-information represented as instrumentation rules.
3) An apparatus for modifying original byte code at runtime as in claim 2, wherein the instrumentation engine obtains a match between a portion of the extracted byte code metadata and a portion of the instrumentation meta-information represented as instrumentation rules.
4) An apparatus for modifying original byte code at runtime as in claim 3, wherein said instrumentation engine inserts metadata extracted from original byte code in an inheritance repository if a match with previous inheritance repository entries cannot be attained.
5) An apparatus for modifying original byte code at runtime as in claim 3, wherein said instrumentation engine updates metadata extracted from original byte code in an inheritance repository if a match with previous inheritance repository entries can be attained.
6) An apparatus for modifying original byte code at runtime as in claim 5, wherein the instrumentation engine incrementally builds a graph within an inheritance repository that maps direct and indirect inheritance and interface implementation relationships using direct inheritance and interface implementation relationships extracted from original byte code received from the agent.
7) An apparatus for modifying original byte code at runtime as in claim 6, wherein the graph that maps direct and indirect inheritance and interface implementation relationships is reused between different application runs.
8) An apparatus for modifying original byte code at runtime as in claim 6, wherein the instrumentation engine filters the portion of the byte code to be modified based upon filter criteria.
9) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria includes a class name.
10) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria includes a method name.
11) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria includes a method argument.
12) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria includes a method modifier.
13) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria include direct and indirect object oriented inheritance and interface implementation relationships of class represented by original byte code.
14) An apparatus for modifying original byte code at runtime as in claim 8, wherein the filter criteria includes a concrete rule.
15) An apparatus for modifying original byte code at runtime as in claim 13, wherein the filter criteria include direct and indirect base classes and implemented interfaces of the class represented by original byte code and wherein the filter criteria are evaluated against direct and indirect inheritance and interface implementation relationships available in the inheritance repository.
16) An apparatus for modifying original byte code at runtime as in claim 13, wherein the filter criteria include directly and indirectly inheriting classes of the class represented by original byte code and wherein the filter criteria are evaluated against direct and indirect inheritance and interface implementation relationships available in the inheritance repository.
17) A method for the Instrumentation of generic component interfaces.
18) A method for Instrumentation of generic component interfaces as in claim 17, wherein concrete implementers of generic component interfaces are selected at runtime.
19) A method for the Instrumentation of generic component interfaces as in claim 17, wherein concrete implementers of generic component interfaces are selected according to direct or indirect implemented interfaces.
20) A method for the Instrumentation of generic component interfaces as in claim 17, wherein concrete implementers of generic component interfaces are selected according to their direct or indirect base classes.
Description
    FIELD OF THE INVENTION
  • [0001]
    The present invention is directed to a system for adaptive and generic bytecode instrumentation for performance monitoring and failure diagnosis tools using load-time or run-time bytecode instrumentation. The placement of instrumentation code is determined at load-time or run-time, taking object-oriented inheritance relationships into account.
  • BACKGROUND OF THE INVENTION
  • [0002]
    The increasing number of features demanded from successful applications and getting shorter development cycles increase complexity of applications and dependency on 3rd party components and libraries.
  • [0003]
    Further, new programming approaches like aspect oriented programming (AOP), which helps mastering source code complexity and thus increase productivity of software development, increase the run-time complexity of applications by adding or altering classes or methods of the application during run-time.
  • [0004]
    Bytecode instrumentation based monitoring and diagnosis systems must provide adaptive, flexible instrumentation placement tools to enable comfortable application monitoring in such complex and dynamic environments.
  • [0005]
    Such instrumentation placement tools must enable instrumentation placement and modification at class load-time or during run-time. Run-time changes of the application code base or 3rd party libraries and components must be managed by the placement tool, e.g. by adding new instrumentation code on-the-fly. The instrumentation tool must cope with different class loading mechanisms, including classes loading from network resources.
  • [0006]
    Further, generic instrumentation placement is required, enabling the instrumentation of top-level component interfaces like e.g. the JDBC driver interfaces, regardless of the concrete implementation of the component. The placement tool should be able to instrument any bytecode based application, regardless of the application type (e.g. J2EE, stand-alone or applet) or underlying bytecode based system (e.g. Sun Microsystems's Java or Microsoft's .NET).
  • [0007]
    Finally, an instrumentation placement tool must not interfere with behavior or functionality of the application by e.g. changing the class load order.
  • DESCRIPTION OF RELATED ART
  • [0008]
    There are several approaches to filter positions for code instrumentations. One approach, as described in [1] and [2], is to specify methods that are target for instrumentation by explicitly specifying the class that contains the method and the method itself. This approach enables basic run-time adaptable instrumentation if instrumentation code is placed near to run-time, e.g. at class loading time, but it lacks information about the internal inheritance structure of the application. To select classes and methods for instrumentation, a user requires in-depth knowledge of the internal structure of the application and used 3rd party components and libraries, including names of classes and methods. In most cases, information at this fine grained level is difficult to acquire for an application and not available for 3rd party components. The Test and Performance Tools Platform (TPTP), a plug-in to the well-known Eclipse development environment, applies this approach and enhances it by enabling the usage of search patterns for the specification of classes and methods. Although search patterns ease the placement of instrumentations because exact knowledge of classes and methods is no longer required, still in-depth knowledge of the application structure is required due to missing information about the internal inheritance structure of the application.
  • [0009]
    An enhanced version of this approach, as described in [3], uses additional meta-data that is available at load-time of class bytecode, like the direct base class or direct implemented interfaces of the loaded class. The meta-data is used to identify methods that should be instrumented, enabling filter criteria taking direct base class or direct implemented interfaces into account. Although this approach enables more generic method filtering, due to the restriction to direct relationships between classes and interfaces, it also lacks information about global inheritance relationships and thus still requires in-depth knowledge of the internal structure of the monitored application.
  • [0010]
    AOP based approaches employing a load-time aspect weaver, which enables altering class bytecode during class loading, use meta-data extracted from bytecode to identify classes and methods. But due to the restricted structural information available at load-time, this approach is also limited to the direct neighborhood of the loaded class, like the direct base class or directly implemented interfaces of the loaded class. Other approaches analyze application deployment data and partially decompile application code to acquire structural information about the application. The system and method presented in [4] uses this approach. The gathered structural information may be used for instrumentation placement. This technique provides fine grained insight into the internal inheritance, but due to the analyzing step which has to be performed prior to run-time, it lacks adaptability to run-time altered application bytecode. Further, every change of the inheritance structure requires a new analysis and decompilation run.
  • [0011]
    Consequently, a need for an alternative approach exists that overcomes the shortcomings of the present approaches.
  • APPENDIX: REFERENCED PATENTS
  • [0012]
    [1] Berry R. F. and Hussain R. Y.; “System and Method for Dynamic Modification of Class Files”; International Business Machines Corporation; US006026234A
  • [0013]
    [2] Cohen G. A. and King R. A.; “Apparatus and Method for Dynamically Modifying Class Files during Loading for Execution”; International Business Machines Corporation; US006026237A
  • [0014]
    [3] Boykin J. R., Giammaria A., Schlosser B. J. and Tapperson K. G.; “Method and System for Auto-Instrumenting Java Applications Through Probe Injection”; 20040123279A1
  • [0015]
    [4] Fenion M. G., Markis A. P. and LaFrance P. J.; “Method and System for Monitoring Distributed Systems”; Diring Software; 20040039728A1
  • SUMMARY OF THE INVENTION
  • [0016]
    The present invention is dedicated to a system and method for adaptive, generic bytecode instrumentation for performance monitoring and failure diagnosis tools using load-time or run-time bytecode instrumentation. Information concerning object-oriented inheritance structures is used to determine methods that are target for instrumentation. A mapping of said inheritance structures is stored in an inheritance repository.
  • [0017]
    The present invention requires neither source code modifications of monitored source code nor access to the source code to instrument bytecode based software applications. The code of monitored applications is instrumented on the fly, during application run-time, during or after loading class bytecode.
  • [0018]
    The present invention provides generic, rule based instrumentation filters to select methods of the application for instrumentation. The rule based instrumentation filters use inheritance information for method selection.
  • [0019]
    Said inheritance information is generated during application run-time and dynamically adapted to changes of configuration or bytecode of the monitored application.
  • [0020]
    The generic filter mechanism enables placement rules that refer to well known, standardized high level component interfaces. The concrete code that implements the functionality of those components is identified and instrumented at run-time.
  • [0021]
    Capturing of inheritance information and placement of instrumentation code has no impact on class loading order.
  • [0022]
    An agent is injected into the process of a monitored application during startup. The agent initializes a communication link with an instrumentation server and intercepts class load events. The bytecode of loaded classes is sent to the instrumentation server for instrumentation. Additionally, instrumentation code can be injected to already loaded classes by sending corresponding bytecode to the instrumentation server and replacing original bytecode with instrumented bytecode in the runtime environment.
  • [0023]
    The instrumentation engine uses meta-data embedded in the received bytecode, like name of the class, direct base class and directly implemented interfaces to incrementally build the inheritance repository that mirrors the inheritance structure of the monitored application. The inheritance repository reveals also indirect class relationships like inheritance relations that span multiple levels.
  • [0024]
    After the inheritance repository is updated with the meta-data of the received bytecode, the instrumentation engine uses the inheritance repository to evaluate instrumentation filter rules to filter methods for instrumentation.
  • [0025]
    The instrumentation engine adds instrumentation code to the code of the filtered methods and sends the altered bytecode back to the agent.
  • [0026]
    The agent forwards the received bytecode to the run-time system which loads the altered bytecode into memory.
  • [0027]
    The present invention enables the instrumentation of generic component interfaces by dynamically selecting concrete implementers of the generic component interfaces for the instrumentation at runtime, the instrumentation of interfaces including selecting the classes that implement the methods to instrument by specifying direct or indirect implemented interfaces and a instrumentation of abstract methods including selecting classes that implement methods to instrument by specifying the direct or indirect base class.
  • [0028]
    The present invention uses object oriented inheritance information to select the methods for instrumentation including creating inheritance repository that maps the global inheritance structure of a monitored application including extracting meta-information identifying direct base class from java or .net class bytecode, storing direct inheritance relationship in a repository, extracting meta-information identifying direct implemented interfaces from java or .net class bytecode, storing the direct interface implementation relationship in a repository and incrementally combining the direct inheritance relationships to map the indirect inheritance relationships of the classes and the interfaces.
  • [0029]
    Additionally, the present invention evaluates the instrumentation rules, using the global inheritance information at class load time to filter methods to instrument, including using the global inheritance information of the class that implements the method, using the global inheritance information of type of one or more method arguments, using the global inheritance information of a type of method return value and using a combination of the above.
  • [0030]
    The instrumentation rule matches classes directly or indirectly extending a specific class, and the instrumentation rule matches methods implementing an interface contract of a specific the directly or indirectly implement interface. The instrumentation rule matching is directly or indirectly inherited but not redefined methods.
  • [0031]
    The present invention adapts object oriented inheritance information according to the changes of monitored application, adapts the inheritance information to load time changed inheritance relationships (AOP load time weaving) including inserting or removing from a base class from the inheritance hierarchy of a class, inserting or removing a base interface from the inheritance hierarchy of an interface and inserting or removing interfaces a class implements, adapts to changes of class definition at load time or during runtime (AOP load time weaving) including inserting or removing methods and changing method signatures.
  • [0032]
    The present invention adapts the inheritance information to runtime added classes and interfaces including the classes loaded from network sources and including classes dynamically defined at runtime. The present invention reuses the inheritance information between application runs and adapts inheritance information acquired from the previous run to changed inheritance relationships caused by the changed application configuration.
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • [0033]
    FIG. 1 illustrates an instrumentation program using instrumentation meta-information and class inheritance-information to evaluate filter rules to determine methods that need to be augmented with instrumentation operations.
  • [0034]
    FIG. 2 illustrates a system, a preferable embodiment for instrumenting the original bytecode at load-time or during run-time.
  • [0035]
    FIG. 3 provides a flowchart for the process of initializing or loading an inheritance repository on agent registration.
  • [0036]
    FIG. 4 depicts an inheritance repository node that may be created for each class or interface received in bytecode format.
  • [0037]
    FIG. 5 provides a flowchart for the process of building and updating an inheritance repository according to class inheritance information extracted from received bytecode.
  • [0038]
    FIG. 6 a) depicts the state of an exemplary inheritance structure stored within the inheritance repository before a specific class is loaded.
  • [0039]
    FIG. 6 b) shows the state of the exemplary inheritance structure after the specific class is loaded.
  • [0040]
    FIG. 7 a) provides a flowchart describing the process of storing the inheritance repository of a disconnecting agent.
  • [0041]
    FIG. 7 b) shows the process of saving inheritance repositories of connected agents on shutting down the instrumentation server.
  • [0042]
    FIG. 8 provides a tabular overview of selected instrumentation rule types enabled by the present invention.
  • [0043]
    FIG. 9 a) shows a flowchart describing the evaluation of concrete instrumentation rules on received class bytecode.
  • [0044]
    FIG. 9 b) illustrates the process of evaluating generic downward rules on received class bytecode.
  • [0045]
    FIG. 9 c) provides a flowchart describing the evaluation of generic upward rules on received class byte code.
  • DESCRIPTION OF EXEMPLARY EMBODIMENTS
  • [0046]
    Bytecode based software platforms like Sun Microsystems's Java or Microsoft's .NET framework provide interfaces to intercept bytecode loading events and to alter the bytecode of the loaded class before it is loaded into memory. Additionally those frameworks enable restricted altering of bytecode that is already loaded into the runtime environment.
  • [0047]
    Open and documented bytecode formats enable analysis of class byte code and selective altering of methods. Meta-data contained in the loaded bytecode enables retrieval of the class name, package or namespace membership, name and signature of implemented methods, direct base class, direct implemented interfaces, etc. The isolated meta-data of a class only reveals direct relationships to other classes, which is not sufficient to provide rule-based adaptive method filtering.
  • [0048]
    The present invention discloses a method to incrementally aggregate isolated class meta-data into a repository that mirrors also indirect relationships between classes and interfaces. This inheritance repository is in turn used to evaluate method filtering rules to detect methods for instrumentation.
  • [0049]
    FIG. 1 shows an exemplary method and system for configuring an instrumentation engine 101, using instrumentation meta-information in the form of instrumentation rules 102, instrumentation operations 104 and, run-time generated application meta-information stored in an inheritance repository 103.
  • [0050]
    Instrumentation rules may either be concrete 812 or generic 814. Parameters of concrete rules include but are not limited to a class name and a method name. Both method and class names may either be concrete names that match to a specific method of a specific class, or name patterns that match a set of methods and/or a set of classes.
  • [0051]
    Generic rules additionally include a direction indicator 820 that specifies how inheritance relationships should be used for rule evaluation.
  • [0052]
    Both concrete and generic rules are described in detail in FIG. 8.
  • [0053]
    The inheritance repository 103 includes run-time created information 112 about class inheritance and interface implementation. The repository is also adapted to changes of inheritance relations during run-time. The process of creation and adaptation of the inheritance repository during run-time is depicted in FIG. 5. The inheritance repository 103 is used to parameterize generic rules which are evaluated during application run-time.
  • [0054]
    The instrumentation operations 104 define the functionality that is injected into instrumented methods. The instrumentation operations 104 preferably comprise bytecode targeted to the application platform. A separate mapping is maintained that associates instrumentation rules with instrumentation operations to enable rule specific instrumentation.
  • [0055]
    The instrumentation engine 101 creates and adapts the inheritance repository 103 during runtime, which is used to parameterize instrumentation rules 102. Parameterized instrumentation rules 102 identify the methods where instrumentations should be placed. The instrumentation operations 104 what instrumentation code is placed.
  • [0056]
    FIG. 2 displays a system 200 including an exemplary embodiment of the present invention consisting in a native loader 204, an agent 205, and an instrumentation server 201 with an instrumentation engine and an inheritance repository. The system also includes an application 207 which is monitored by the embodiment and the runtime environment 206 of the application 207. Agent 205 and native loader 204 are deployed to the application 207. Application 207 and instrumentation server 201 may be separated to different hosts and use a network connection for communication and to exchange bytecode, or they may run on the same host, using inter-process communication to exchange messages and bytecode. Another embodiment may run the instrumentation server 201 within the process of the application 207.
  • [0057]
    An agent 205 is injected into the process of the application during the application start and establishes a communication link to the instrumentation server 201. The instrumentation server 201 can handle multiple connected agents simultaneously.
  • [0058]
    Instrumenting a software application 207 at load time comprises the following steps: intercepting the native class loader 204 using an agent 205; capturing the original bytecode 202 and transferring said original bytecode 202 to an instrumentation engine 101; updating the inheritance repository 103 shown in FIG. 1 with meta-data extracted from original bytecode 202; evaluating instrumentation rules 102 by using the inheritance repository 103 to filter methods, and instrumenting bytecode of filtered methods; returning the instrumented bytecode 203 to the agent 205, which in turn forwards the instrumented bytecode 203 to the application run-time module 206 instead of the original bytecode 202.
  • [0059]
    Alternatively, a software application can be instrumented at run-time which allows altering the set of deployed instrumentations during run-time, without the need for an application restart. Instrumenting a software application 207 at run-time comprises the following steps: original bytecode 202 is fetched either by the agent 205 from the application run-time module 206, or retrieved from a bytecode cache of the instrumentation server 201; in case of bytecode fetched from the application run-time, the inheritance repository 103 is updated with meta-data extracted from original bytecode 202; evaluating instrumentation rules 102 and instrument original bytecode 202; returning instrumented bytecode 203 to the agent 205, which replaces the original bytecode 202 of the application run-time module 206 by the instrumented bytecode 203.
  • [0060]
    Referring to FIG. 3, in step 306 a new inheritance repository 103 is created, initialized and assigned to an agent 205 that registers at instrumentation server 101 for the first time. In step 304, a determination is made to find if an existing inheritance repository is available. For subsequent registrations, the inheritance repository 103 assigned to the agent is loaded step 308. After the inheritance repository 103 for the agent 205 is loaded, the agent establishes a connection to the instrumentation engine 101 in step 310.
  • [0061]
    FIG. 4 depicts the preferred embodiment of an inheritance repository node (IRN) 401, which is used to represent classes or interfaces within the inheritance repository. It includes meta-information representing a class or interface received from the agent 205, and references to other IRNs 401 which map inheritance or interface implementation relationships. An IRN 401 includes the class name 402 of the represented class or interface, a flag indicating if the IRN describes an interface or a class 403, and a list of declared methods 404 containing method related meta-information like method name and signature.
  • [0062]
    The field super class 405 references to the IRN 401 representing the direct super class or super interface of the described class or interface and the field implemented interfaces 406 is a list of references to IRNs representing interfaces directly implemented by the class described by the IRN. Classes that directly extend the described class are identified by the field inheriting classes 407, which is a list of references to IRNs representing the classes directly extending the described class.
  • [0063]
    IRNs and references between IRNs build a graph that enables queries for direct and indirect inheritance and interface implementation relations between classes.
  • [0064]
    FIG. 5 illustrates the process of building and updating the inheritance repository 103 according to meta-information extracted from original bytecode 202 received from the agent 205.
  • [0065]
    In a first step 502, meta-information like class or interface name, direct super class name and names of directly implemented interfaces are extracted from the received original bytecode 202. The extracted class or interface name is used to query the inheritance repository 103 for an existing IRN 401 that represents the received class. In step 504, it is determined if the inheritance repository already contains an IRN representing the received original bytecode. If a matching IRN is found, the meta-information of the IRN 401 is updated with the meta-information extracted from received original bytecode 202 in step 506. Otherwise, a new IRN 401 is initialized with meta-information from original bytecode 202 and added to the inheritance repository 103 in step 508.
  • [0066]
    The inheritance repository is queried in step 510 for IRNs representing the direct super class and the directly implemented interfaces extracted from original bytecode 202, and new IRNs are created in step 512 for classes and interfaces that are not represented within the inheritance repository 103. Those IRNs are partially initialized in step 512 by setting the class name 402 and the interface flag 403 with the meta-information extracted from original bytecode 202. The remaining part of the IRNs created in step 512 is initialized when bytecode of the represented classes is loaded and processed by the instrumentation engine 101. Further, the IRN 401 representing the super class of the received class or interface is updated in step 514 by adding a reference to the IRN representing the received original bytecode 202 to the list of inheriting classes 407.
  • [0067]
    In a subsequent part of step 514, the fields super class 405, and implemented interfaces 406 of the IRN 401 representing the received original bytecode 202 are initialized with references to the IRNs representing the direct super class and the directly implemented interfaces of the class represented by the received original bytecode 202.
  • [0068]
    The process depicted in FIG. 5 incrementally builds a graph that maps global inheritance and interface implementation relationships of all classes and interfaces received from the agent 205.
  • [0069]
    FIG. 6 exemplary illustrates the process of updating an inheritance repository 103. FIG. 6 a) shows the state of the inheritance repository before inserting meta-data extracted from received original bytecode of class A. FIG. 6 b) shows the state of the inheritance repository after the IRN representing class A was added to the repository. Prior to inserting meta-information of class A, the inheritance repository contains separated IRN 401 graphs, describing parts of the inheritance structure. One sub graph 620 maps inheritance relationship from class F 607 to class E 605 and class D 606, and the interface implementation relationship from class D 606 to interface I 608. Another graph 622 maps the inheritance relationship between the interfaces K 610 and J 609. A third graph 624 contains IRNs for the classes B 603 and C 604 and a preliminary IRN of class A 601. The preliminary IRN of class A 601 contains meta-information concerning class A that was extracted from original bytecode 202 of classes B 603 and C 604, indicating a common super class A 601.
  • [0070]
    On receiving original bytecode 202 representing class A, the preliminary IRN of class A 601 is updated with extracted meta-information to the final IRN of class A 602. Meta-information extracted from original bytecode 202 representing class A 602 reveals an inheritance relationship between class D 606 and class A 602 and an interface implementation relationship between class A 602 and interface J 609. The inheritance repository 103 is updated to map these additional identified relationships, which fills the gap within the inheritance repository 103 and connects the separated graphs.
  • [0071]
    The graph depicted in FIG. 6 b) shows direct and indirect inheritance relationships and interface implementation relation ships of class A 602. For instance, class A 602 directly inherits from class D 606 because both classes are directly connected in the inheritance graph, and it indirectly inherits from class F 607 because class A 602 and class F 607 are indirectly connected via class D 606. Additionally direct and indirect interface relationships of Class A 602 are shown. As an example class A 602 directly implements interface J 609, because the class is directly connected to the interface. Said class A 602 indirectly implements interface K 610 because interface K 610 and class A 602 are indirectly connected via interface J 609.
  • [0072]
    FIG. 7 describes storage of inheritance repositories 103 by a preferred embodiment using a separate instrumentation server 201 handling multiple agent connections. As depicted in FIG. 7 a), the inheritance repository 103 associated with a specific agent 205 is stored in step 702 if the agent is disconnected in step 704. As depicted in FIG. 7 b), inheritance repositories 103 of all connected agents are stored in step 706 on shutdown in step 708 of the instrumentation server 201. The stored inheritance repositories are used on subsequent agent connections (cf. FIG. 3).
  • [0073]
    The preferred embodiment evaluates instrumentation rules 102 against the name of the class or interface and the method names extracted from received original bytecode 202, to filter methods for instrumentation. Instrumentation rules are grouped into concrete and generic rules. Referring now to FIG. 8, concrete instrumentation rules 812 provide filter criteria 810 for class name and a method name which are matched against class name and method names extracted from original bytecode 202. Both class and method filter criteria 810 may be search patterns that match to multiple extracted class or method names. Additional filter criteria 810, like method arguments and method modifiers may also be used for filtering.
  • [0074]
    Generic rules 814 additionally take inheritance relationships into account. The class name filter criterion is used to query the inheritance repository 103 for classes or interfaces with specific inheritance relationships. Generic instrumentation rules 814 enable instrumentation of method declarations without executable code, like methods declared in interfaces or abstract methods. The classes implementing those methods are determined and instrumented at run-time.
  • [0075]
    The preferred embodiment provides two different types of generic rules 814, called downward rules 816 and upward rules 818. Generic rules 814 contain a direction indicator 820 which specifies the usage of inheritance information of a specific generic rule. Downward rules 816 are scanning for classes that directly or indirectly inherit from classes, or directly or indirectly implement interfaces matching the class filter criterion. Upward rules are scanning for classes and interfaces that are direct or indirect super classes of the classes matching the class filter criterion.
  • [0076]
    Downward rules 816 enable instrumentation in an application independent way. For instance, downward rules can be used to instrument any JDBC driver, regardless of the internal implementation of the driver.
  • [0077]
    Upward rules 818 are used to instrument methods inherited but not overwritten by classes matching the class selection criteria.
  • [0078]
    The information of the inheritance repository 103 enables various other types of generic instrumentation rules, like rules filtering only methods that implement interfaces, or rules filtering methods called by a specific method.
  • [0079]
    FIG. 8 exemplary shows the evaluation of concrete, upward, and downward rules. Column 801 shows the filtering of method M1 of class Z using a concrete instrumentation rule with a class filter criterion “Z” and method filter criterion “M1” or “*”, a wildcard matching any method name and thus selecting any method of class Z. Generic downward filtering rules are demonstrated in column 802. Class Z implements interface A, which declares M1. In turn, class Z implements method M1 to fulfill the contract of interface A. A downward rule 816 specified for interface A, selecting declared method M1 or all declared methods implicitly filters M1 of Z by evaluating the interface implementation relationship between Z and A. Additionally, class Z extends class B and overwrites the method M2 declared by class B. A downward rule specified for class B, selecting method M2, or a downward rule selecting all methods of class B implicitly filters method M2 of class Z by evaluating the inheritance relationship between class Z and B.
  • [0080]
    Column 803 illustrates the evaluation of a generic upward rule. Class X extends class Z, and is not overwriting method M1. An upward rule 818 defined for the method M1 or all methods of class X selects method M1 of class Z by searching the nearest super class of X implementing the method M1.
  • [0081]
    Referring to FIG. 9, instrumenting received original bytecode 202 during load-time or run-time is performed by evaluating instrumentation rules to filter methods for instrumentation.
  • [0082]
    FIG. 9 a) depicts the evaluation of concrete instrumentation rules. Meta-information previously extracted from original byte code 205, like class name and method names is used by the instrumentation engine 101 to evaluate instrumentation rules 103. To determine if a method is selected by one or more concrete instrumentation rules the instrumentation engine 101 first selects in step 902 all concrete instrumentation rules matching the extracted method name. In a second step 904, the extracted class name is matched with the class selection criterion of the instrumentation rules selected in the previous step. Instrumentation operations 104 associated with concrete instrumentation rules that passed both matching steps are inserted in step 906 into the received original bytecode 202.
  • [0083]
    The process of evaluating downward rules for a method name extracted from received original bytecode 202 is shown in FIG. 9 b). First, all downward rules matching the extracted method name are selected in step 912. Then, the inheritance repository 103 is queried in step 914 to determine if the class selection criterion of the selected rules match to the name of a direct or indirect super class or a directly or indirectly implemented interface of the class represented by the received original bytecode 202. If matching classes are found in step 916, downward rules with matching class names are selected, and instrumentation operations 104 associated with those rules are inserted into the received original bytecode 202 in step 918.
  • [0084]
    FIG. 9 c) shows the process of evaluating upward rules on a method name extracted from received original bytecode 202. After selecting all upward rules matching the extracted method name in step 922, the inheritance repository 103 is queried for the classes inheriting from the class represented by the received bytecode that match the class selection criterion of the selected rules in step 924. If matching classes are found in step 926, upward rules with class names that match one of the inheriting class names are selected and instrumentation operations 104 associated with those rules are inserted into received original bytecode 202 in step 928.
  • [0085]
    The processes described in FIG. 9 a) to 9 c) are executed for each method name extracted from received original bytecode 202 to generate the instrumented bytecode 203.
Citations de brevets
Brevet cité Date de dépôt Date de publication Déposant Titre
US5432932 *23 oct. 199211 juil. 1995International Business Machines CorporationSystem and method for dynamically controlling remote processes from a performance monitor
US5727147 *8 déc. 199510 mars 1998Sun Microsystems, Inc.System and method for resolving symbolic references to externally located program files
US5781778 *5 juin 199714 juil. 1998International Business Machines CorporationMethod and system for debugging parallel and distributed applications
US5794046 *5 juin 199711 août 1998International Business Machines CorporationMethod and system for debugging parallel and distributed applications
US5867712 *5 déc. 19972 févr. 1999Shaw; Venson M.Single chip integrated circuit system architecture for document instruction set computing
US5933639 *17 mai 19963 août 1999International Business Machines CorporationSystem and method for debugging distributed programs
US5953530 *25 nov. 199714 sept. 1999Sun Microsystems, Inc.Method and apparatus for run-time memory access checking and memory leak detection of a multi-threaded program
US6101524 *23 oct. 19978 août 2000International Business Machines CorporationDeterministic replay of multithreaded applications
US6102966 *20 mars 199815 août 2000Preemptive Solutions, Inc.Method for renaming identifiers of a computer program
US6134603 *20 mars 199817 oct. 2000Sun Microsystems, Inc.Method and system for deterministic hashes to identify remote methods
US6145121 *17 avr. 19987 nov. 2000University Of WashingtonTrace based method for the analysis, benchmarking and tuning of object oriented databases and applications
US6151639 *19 juin 199721 nov. 2000Sun Microsystems, Inc.System and method for remote object invocation
US6202199 *30 juil. 199813 mars 2001Mutek Solutions, Ltd.System and method for remotely analyzing the execution of computer programs
US6266805 *27 juil. 199824 juil. 2001British Telecommunications PlcVisualization in a modular software system
US6332212 *2 oct. 199718 déc. 2001Ltx CorporationCapturing and displaying computer program execution timing
US6539541 *20 août 199925 mars 2003Intel CorporationMethod of constructing and unrolling speculatively counted loops
US6721941 *22 août 200013 avr. 2004Compuware CorporationCollection of timing and coverage data through a debugging interface
US6754890 *14 déc. 199922 juin 2004International Business Machines CorporationMethod and system for using process identifier in output file names for associating profiling data with multiple sources of profiling data
US6760903 *22 août 20006 juil. 2004Compuware CorporationCoordinated application monitoring in a distributed computing environment
US6795962 *30 déc. 199921 sept. 2004Microsoft CorporationMachine independent debugger
US6862711 *28 oct. 19991 mars 2005International Business Machines CorporationMethod and apparatus in a data processing system for providing an interface for non-intrusive observable debugging, tracing, and logging data from execution of an application
US6961926 *13 déc. 20001 nov. 2005Nec CorporationSystem and method for distributed debugging and recording medium on which control programs are recorded
US6968540 *12 juin 200122 nov. 2005Opnet Technologies Inc.Software instrumentation method and apparatus
US6978444 *1 août 200020 déc. 2005International Business Machines CorporationComputer-implemented method and system for automatically invoking a predetermined debugger command at a desired location of a single thread of a program
US7143392 *11 févr. 200228 nov. 2006Hewlett-Packard Development Company, L.P.Hyperbolic tree space display of computer system monitoring and analysis data
US7162710 *1 nov. 20019 janv. 2007Microsoft CorporationDynamic modifications to a heterogeneous program in a distributed environment
US7263689 *1 nov. 200128 août 2007Microsoft CorporationApplication program interface for dynamic instrumentation of a heterogeneous program in a distributed environment
US7293259 *2 sept. 20036 nov. 2007Sun Microsystems, Inc.Dynamically configuring selected methods for instrument-based profiling at application run-time
US7293260 *26 sept. 20036 nov. 2007Sun Microsystems, Inc.Configuring methods that are likely to be executed for instrument-based profiling at application run-time
US7367025 *30 déc. 200329 avr. 2008Sap AgByte code modification for testing, debugging and/or monitoring of virtual machine based software
US7376940 *4 juin 200420 mai 2008Sun Microsystems, Inc.Thread suspension and method in a multi-threaded environment
US7380239 *31 mai 200127 mai 2008Oracle International CorporationMethod and mechanism for diagnosing computer applications using traces
US7409676 *20 oct. 20035 août 2008International Business Machines CorporationSystems, methods and computer programs for determining dependencies between logical components in a data processing system or network
US7493607 *9 juil. 200217 févr. 2009Bluerisc Inc.Statically speculative compilation and execution
US7496903 *12 août 200324 févr. 2009Hewlett-Packard Development Company, L.P.Synthesizing application response measurement (ARM) instrumentation
US7500227 *22 juin 20043 mars 2009Unisys CorporationProcess and system for intercepting a .NET method at runtime
US7526760 *5 nov. 200428 avr. 2009Sun Microsystems, Inc.Methods for implementing virtual method invocation with shared code
US7685183 *4 sept. 200123 mars 2010OP40, IncSystem and method for synchronizing assets on multi-tiered networks
US7818721 *1 févr. 200619 oct. 2010Oracle America, Inc.Dynamic application tracing in virtual machine environments
US7950004 *16 oct. 200624 mai 2011Siemens CorporationDevices systems and methods for testing software
US20010004766 *13 déc. 200021 juin 2001Nec CorporationSystem and method for distributed debugging and recording medium on which control programs are recorded
US20020032754 *4 avr. 200114 mars 2002Gary LogstonMethod and apparatus for profiling in a distributed application environment
US20020174415 *19 juin 200121 nov. 2002Hines Kenneth J.System and method for debugging distributed software environments
US20020199173 *29 janv. 200126 déc. 2002Matt BowenSystem, method and article of manufacture for a debugger capable of operating across multiple threads and lock domains
US20030056200 *19 sept. 200120 mars 2003Jun LiRuntime monitoring in component-based systems
US20040010570 *9 juil. 200215 janv. 2004Kaler Christopher G.Profiling and tracing distributed applications
US20040093588 *12 nov. 200213 mai 2004Thomas GschwindInstrumenting a software application that includes distributed object technology
US20050039171 *12 août 200317 févr. 2005Avakian Arra E.Using interceptors and out-of-band data to monitor the performance of Java 2 enterprise edition (J2EE) applications
US20050039172 *12 août 200317 févr. 2005Jeffrey ReesSynthesizing application response measurement (ARM) instrumentation
US20050039186 *12 août 200317 févr. 2005Borkan Martha S.Use of thread-local storage to propagate application context in Java 2 enterprise editon (J2EE) applications
US20050039187 *12 août 200317 févr. 2005Avakian Arra E.Instrumenting java code by modifying bytecodes
US20050039190 *12 août 200317 févr. 2005Jeffrey ReesPropagating web transaction context into common object model (COM) business logic components
US20050086656 *20 oct. 200321 avr. 2005Gemstone Systems, Inc.Methods and systems for inter-process copy sharing of data objects
US20050223367 *29 mars 20056 oct. 2005Tonic Solutions, Inc.System and methods for instrumenting applications
US20050278706 *24 mai 200515 déc. 2005International Business Machines CorporationSystem, method, and computer program product for logging diagnostic information
US20060069682 *16 sept. 200430 mars 2006Microsoft CorporationProfiler tool for database servers
US20060271395 *25 mai 200630 nov. 2006Harris Steven TDistributed object identity in a virtual machine cluster
US20060271542 *25 mai 200630 nov. 2006Harris Steven TClustered object state using logical actions
US20060271575 *25 mai 200630 nov. 2006Harris Steven TClustered object state using field set operations
US20060271930 *25 mai 200630 nov. 2006Letizi Orion DClustered object state using synthetic transactions
US20060271931 *25 mai 200630 nov. 2006Harris Steven TDistributed signaling in a virtual machine cluster
US20070011667 *25 mai 200611 janv. 2007Saravanan SubbiahLock management for clustered virtual machines
US20070069005 *29 sept. 200529 mars 2007Dickerson Scott SMethod and system for identifying unsafe synthetic transactions and modifying parameters for automated playback
US20070088762 *25 mai 200619 avr. 2007Harris Steven TClustering server providing virtual machine data sharing
US20070143323 *15 déc. 200521 juin 2007Wily Technology, Inc.Correlating cross process and cross thread execution flows in an application manager
US20070143743 *15 déc. 200521 juin 2007Wily Technology, Inc.Use of execution flow shape to allow aggregate data reporting with full context in an application manager
US20070169055 *11 août 200619 juil. 2007Bernd GreifenederMethod and system for automated analysis of the performance of remote method invocations in multi-tier applications using bytecode instrumentation
US20070180439 *1 févr. 20062 août 2007Sun Microsystems, Inc.Dynamic application tracing in virtual machine environments
Citations hors brevets
Référence
1 *web article (Metadata API Overview, dated May 27, 2006, URL: http://msdn.microsoft.com/en-us/library/ms404434)
Référencé par
Brevet citant Date de dépôt Date de publication Déposant Titre
US8307345 *4 nov. 20086 nov. 2012Ca, Inc.Intelligent engine for dynamic and rule based instrumentation of software
US8387023 *16 oct. 200826 févr. 2013International Business Machines CorporationInstrumentation of computer-based application software
US84642256 mai 200711 juin 2013Dynatrace Software GmbhMethod and system for adaptive, generic code instrumentation using run-time or load-time generated inheritance information for diagnosis and monitoring application performance and failure
US847392511 mai 201025 juin 2013Ca, Inc.Conditional dynamic instrumentation of software in a specified transaction context
US849929929 juin 201030 juil. 2013Ca, Inc.Ensuring deterministic thread context switching in virtual machine applications
US853368730 nov. 200910 sept. 2013dynaTrade Software GmbHMethods and system for global real-time transaction tracing
US856680011 mai 201022 oct. 2013Ca, Inc.Detection of method calls to streamline diagnosis of custom code through dynamic instrumentation
US858989428 févr. 201219 nov. 2013International Business Machines CorporationSelf-monitoring object-oriented applications
US8612573 *28 août 200817 déc. 2013Ca, Inc.Automatic and dynamic detection of anomalous transactions
US873267029 juin 201020 mai 2014Ca, Inc.Ensuring determinism during programmatic replay in a virtual machine
US8745598 *14 déc. 20103 juin 2014Bmc Software, Inc.Running injected code prior to execution of an application
US8752015 *5 déc. 201110 juin 2014Ca, Inc.Metadata merging in agent configuration files
US8752026 *1 juin 200910 juin 2014International Business Machines CorporationEfficient code instrumentation
US876950320 août 20071 juil. 2014The Mathworks, Inc.Computation of elementwise expression in parallel
US8769518 *29 juin 20101 juil. 2014Ca, Inc.Ensuring determinism during programmatic replay in a virtual machine
US878261211 mai 201015 juil. 2014Ca, Inc.Failsafe mechanism for dynamic instrumentation of software using callbacks
US8799871 *8 janv. 20075 août 2014The Mathworks, Inc.Computation of elementwise expression in parallel
US880643729 juin 201112 août 2014International Business Machines CorporationAutomated testing process
US8863093 *6 mars 200914 oct. 2014Coverity, Inc.Load-time instrumentation of virtual machine program code
US887510630 mars 201228 oct. 2014International Business Machines CorporationAutomated testing process
US893872912 oct. 201020 janv. 2015Ca, Inc.Two pass automated application instrumentation
US9015668 *10 août 201121 avr. 2015Interactive TKO, Inc.Instrumentation agent for manipulating component responses in a test
US904741225 févr. 20132 juin 2015Dynatrace CorporationSystem and method for extracting instrumentation relevant inheritance relationships for a distributed, inheritance rule based instrumentation system
US9141415 *2 nov. 201122 sept. 2015Syddansk UniversitetMethod for dynamically transforming the bytecode of Java virtual machine bootstrap classes
US9170825 *21 avr. 201127 oct. 2015Oracle International CorporationInterface method resolution for virtual extension methods
US923185817 déc. 20105 janv. 2016Dynatrace Software GmbhCompleteness detection of monitored globally distributed synchronous and asynchronous transactions
US9250925 *13 avr. 20102 févr. 2016Sybase, Inc.Adding inheritance support to a computer programming language
US927491925 avr. 20121 mars 2016Dynatrace Software GmbhTransaction tracing mechanism of distributed heterogenous transactions having instrumented byte code with constant memory consumption and independent of instrumented method call depth
US94116169 déc. 20119 août 2016Ca, Inc.Classloader/instrumentation approach for invoking non-bound libraries
US9411964 *24 nov. 20149 août 2016Bluerisc, Inc.Characterizing, detecting and healing vulnerabilities in computer code
US9542210 *2 juin 201410 janv. 2017Ca, Inc.Ensuring determinism during programmatic replay in a virtual machine
US956371914 oct. 20137 févr. 2017International Business Machines CorporationSelf-monitoring object-oriented applications
US960682022 avr. 201428 mars 2017Ca, Inc.Ensuring determinism during programmatic replay in a virtual machine
US970368721 sept. 201211 juil. 2017Hewlett Packard Enterprise Development LpMonitor usable with continuous deployment
US975411224 nov. 20145 sept. 2017Bluerisc, Inc.Detection and healing of vulnerabilities in computer code
US9798557 *24 août 201224 oct. 2017Ca, Inc.Injection of updated classes for a java agent
US981136229 févr. 20167 nov. 2017Dynatrace Software GmbhMethod and system for transaction controlled sampling of distributed heterogeneous transactions without source code modifications
US981765624 août 201214 nov. 2017Ca, Inc.Hot rollback of updated agent
US20080178165 *8 janv. 200724 juil. 2008The Mathworks, Inc.Computation of elementwise expression in parallel
US20090055806 *22 août 200726 févr. 2009Jian TangTechniques for Employing Aspect Advice Based on an Object State
US20090144747 *20 août 20074 juin 2009The Mathworks, Inc.Computation of elementwise expression in parallel
US20100058345 *28 août 20084 mars 2010David Isaiah SeidmanAutomatic and dynamic detection of anomolous transactions
US20100100873 *16 oct. 200822 avr. 2010Konstantin ShaginInstrumentation of Computer-Based Application Software
US20100115495 *4 nov. 20086 mai 2010Bhaskar SunkaraIntelligent engine for dynamic and rule based instrumentation of software
US20100306745 *1 juin 20092 déc. 2010International Business Machines CorporationEfficient Code Instrumentation
US20110252400 *13 avr. 201013 oct. 2011Sybase, Inc.Adding inheritance support to a computer programming language
US20120151450 *9 déc. 201014 juin 2012Microsoft CorporationPlatform-Agnostic Diagnostic Data Collection and Display
US20120151464 *14 déc. 201014 juin 2012Bmc Software, Inc.Running Injected Code Prior to Execution of an Application
US20120272214 *21 avr. 201125 oct. 2012Oracle International CorporationInterface method resolution for virtual extension methods
US20130145349 *5 déc. 20116 juin 2013Computer Associates Think, Inc.Metadata Merging In Agent Configuration Files
US20140059527 *24 août 201227 févr. 2014Ca, Inc.Injection of updated classes for a java agent
US20140075422 *2 nov. 201113 mars 2014Syddansk UniversitetMethod for dynamically transforming the bytecode of java virtual machine bootstrap classes
US20140282555 *2 juin 201418 sept. 2014Ca, Inc.Ensuring Determinism During Programmatic Replay in a Virtual Machine
US20150160947 *8 déc. 201411 juin 2015Identify Software, Ltd. (IL)Estimating indirect interface implementation before load time based on directly implemented methods
US20160202960 *14 janv. 201614 juil. 2016Dynatrace LlcMethod And System For Automated, Static Instrumentation For Applications Designed For Execution In Environments With Restricted Resources, Like Mobile Devices Or TV Set Top Boxes
CN102650964A *28 févr. 201129 août 2012国际商业机器公司Method and system for monitoring object-oriented application and self-monitoring system
CN104641353A *21 sept. 201220 mai 2015惠普发展公司,有限责任合伙企业Monitor usable with continuous deployment
EP2898411A4 *21 sept. 201220 avr. 2016Hewlett Packard Development CoMonitor usable with continuous deployment
Classifications
Classification aux États-Unis717/130
Classification internationaleG06F9/44
Classification coopérativeG06F11/3624, G06F11/3644
Classification européenneG06F11/36B2, G06F11/36B6
Événements juridiques
DateCodeÉvénementDescription
16 juil. 2010ASAssignment
Owner name: DYNATRACE SOFTWARE GMBH, AUSTRIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GREIFENEDER, BERND;REEL/FRAME:024696/0556
Effective date: 20100715
19 nov. 2013CCCertificate of correction
24 nov. 2016FPAYFee payment
Year of fee payment: 4